In my years of foundry engineering practice, I have worked extensively with different casting processes, and one of the most transformative experiences has been the adoption of the lost foam casting (LFC) process for producing blast furnace cooling walls. This article is a first-hand account of how I led the transition from conventional sand casting methods to the lost foam castings route, the technical challenges we overcame, the process parameters we optimized, and the remarkable economic and quality improvements that followed. Throughout this article, I will repeatedly emphasize the critical role of lost foam castings in modern foundry operations, particularly for complex cored castings such as cooling walls with embedded cooling tubes.
Technical Requirements of the Cooling Wall
Blast furnace cooling walls are classified as embedded-tube castings, which are notoriously difficult to manufacture due to the need to maintain precise dimensional tolerances while accommodating pre-bent steel tubes. The cooling wall I was responsible for had to satisfy strict dimensional and material specifications. The thickness tolerance was ±5 mm, the height tolerance was also ±5 mm, and the chord length tolerance across the inner and outer arcs was ±3 mm. The bosses at the roots of the cooling water inlet and outlet pipes, as well as the bosses at bolt holes, had to be smooth and flat with a height tolerance of ±1.5 mm. All cooling water pipe inlet and outlet center positions had to be within a positional tolerance of ±3 mm. After cleaning, the embedded pipes were subjected to pressure testing: the pressure had to be held for 10 minutes, with the pressure drop not exceeding 3% within the first 5 minutes. The pipe material specified was #10 steel, conforming to GB/T 699–2015. The chemical composition and mechanical property requirements for the cooling wall body material are summarized in Table 1.
| Material grade | C (wt%) | Si (wt%) | Mn (wt%) | P (wt%) | S (wt%) | Tensile strength (MPa) | Elongation (%) |
|---|---|---|---|---|---|---|---|
| HT200 | 3.2–3.5 | 1.6–2.2 | 0.6–0.9 | <0.20 | <0.150 | ≥200 | – |
| QT500-7 | 3.5–4.0 | ≤2.7 | ≤0.6 | ≤0.08 | ≤0.025 | ≥420 | ≥4 |
| QT400-18 | 3.4–4.0 | ≤2.8 | ≤0.5 | ≤0.07 | ≤0.030 | ≥370 | ≥12 |
For HT200, the test bar was a separately cast specimen, while for QT500-7 and QT400-18, the test bars were attached to the casting. These demanding requirements made the process selection crucial. In my opinion, lost foam castings offered a unique opportunity to achieve both dimensional consistency and high surface quality while reducing labor and environmental burdens.
Lost Foam Casting Process Layout
The process flow I used for producing cooling walls through lost foam castings is illustrated in the sequence of operations: foam pattern fabrication, pattern coating, coating drying, sprue assembly, embedding of steel cooling pipes, coating of the assembly, drying, placement in a flask, filling with dry sand, vacuum application, pouring of molten metal, cooling, shakeout, shot blasting, pressure testing, finishing, inspection, and final dispatch. Each step required meticulous control to prevent defects such as coating cracking, pattern deformation, or sand collapse. The foam pattern material was expanded polystyrene (EPS) with a density of approximately 20–22 g/L, which provided sufficient strength and stiffness for the large cooling wall patterns. The pattern was cut and assembled from pre-molded segments to minimize warpage.
A crucial step in our lost foam castings process was the application of the refractory coating. We used a water-based coating containing zirconium silicate and mica, applied in two layers. The first layer had a viscosity of 45–55 s (Ford cup #4), and the second layer was slightly thinner at 35–45 s. The total dry coating thickness was kept at 0.8–1.2 mm. After each coating layer, the pattern was dried in a controlled environment at 45–55 °C for 6–8 hours to thoroughly remove moisture without softening the foam. Insufficient drying led to steam-related defects such as folds and carbon inclusions in the final casting.
For the embedded cooling tubes, we inserted the pre-formed steel tubes into the foam pattern before the final coating application. The tubes were positioned using fixtures to ensure their centerlines aligned with the pattern reference marks. The foam bosses around the tube openings were molded with a density slightly higher (25 g/L) to strengthen the joint area. This attention to detail was necessary because the dimensional accuracy of these boss areas is a common source of rejection in cooling wall castings.
Iron Liquid Treatment and Process Parameters
The metallurgical treatment of the iron liquid played a vital role in achieving the required mechanical properties of the lost foam castings. For ductile iron grades QT500-7 and QT400-18, I performed nodularization using a sandwich method in a treatment ladle. The nodulizer (FeSiMg alloy with 5–8% Mg) was placed in a pocket on one side of the ladle, covered first with a layer of ferrosilicon granules, then with a layer of steel shot, and rammed firmly. The molten iron was carefully tapped into the ladle so that the first two-thirds of the iron liquid struck the opposite side, gradually dissolving the nodulizer. After the nodulization reaction was complete, the remaining one-third of the iron was tapped into the ladle while a 75% ferrosilicon inoculant was added concurrently. For QT500-7, the inoculation addition was 0.4–0.5% of the iron liquid weight; for QT400-18, it was 0.7–1.0%. Additionally, for QT400-18, a post-inoculation with 0.2–0.3% ferrosilicon granules was performed immediately before pouring to enhance graphite nodule count.
Controlling the vacuum during pouring was one of the most critical factors in our lost foam castings operation. The vacuum level had to be optimized for different materials because the gas generation rate and the permeability of the sand compact differ. For gray iron HT200, I maintained the vacuum at 0.04–0.05 MPa. For ductile iron grades, due to the higher pouring temperature and increased gas evolution from the foam decomposition, the vacuum was raised to 0.05–0.06 MPa. This higher vacuum helped to draw the decomposition gases more rapidly through the refractory coating and the sand, minimizing the risk of gas porosity and incomplete fill.
Pouring temperature also had a significant influence on the quality of the lost foam castings. A low pouring temperature could result in cold shots and misruns because the foam decomposition absorbs heat and creates a gas gap; an excessively high temperature can cause sand collapse or metal penetration through the coating. For gray iron, I controlled the pouring temperature at 1,290–1,310 °C. For ductile iron, the recommended range was 1,310–1,320 °C. By carefully adjusting the temperature within these windows, we achieved excellent surface finish and dimensional fidelity in every cooling wall.
The chemical compositions and resulting mechanical properties from actual production runs are presented in Table 2. These values are the average of multiple heats and demonstrate that the lost foam castings consistently met the technical specifications.
| Material | Heat | C (%) | Si (%) | Mn (%) | P (%) | S (%) | Mg residual | RE residual | Tensile strength (MPa) | Elongation (%) |
|---|---|---|---|---|---|---|---|---|---|---|
| HT200 | 1 | 3.44 | 1.83 | 0.64 | 0.083 | 0.044 | – | – | 206 | – |
| HT200 | 2 | 3.48 | 2.01 | 0.75 | 0.081 | 0.040 | – | – | 211 | – |
| QT500-7 | 1 | 3.75 | 2.42 | 0.57 | 0.054 | 0.026 | 0.039 | 0.021 | 481 | 5.5 |
| QT500-7 | 2 | 3.81 | 2.38 | 0.50 | 0.045 | 0.023 | 0.045 | 0.025 | 474 | 5.9 |
| QT400-18 | 1 | 3.59 | 2.67 | 0.43 | 0.049 | 0.028 | 0.051 | 0.032 | 440 | 13.9 |
| QT400-18 | 2 | 3.65 | 2.58 | 0.36 | 0.041 | 0.021 | 0.043 | 0.028 | 456 | 14.6 |
For the ductile iron attached test bars, I examined the graphite morphology under a microscope. The nodularity for both QT500-7 and QT400-18 was rated as 2–3 according to the standard charts, meaning the graphite was predominantly spheroidal with a few degenerate particles. The ferrite content appeared commensurate with the elongation values we recorded. This confirmed that the nodularization and inoculation practices in our lost foam castings were effective.
Dimensional Analysis of Lost Foam Castings Cooling Walls
After the initial production batch, I conducted a comprehensive dimensional inspection on every cooling wall. The results were encouraging but also revealed one recurring deviation. The bolt hole locations, thickness dimensions, and chord length dimensions all met the specified tolerances. However, the width dimension showed an oversize condition in some castings, with the deviation being less than 10 mm. This was particularly noticeable in ductile iron castings rather than gray iron. The cause is rooted in the graphitization expansion during solidification of ductile iron, which can distort the foam pattern cavity if the sand mold is not sufficiently rigid. In lost foam castings, the sand is not bonded, so its compaction and the vacuum level determine how much expansion the mold wall can resist.
To address this, I negotiated with the client to allow a tolerable oversize of 5 mm without rework. For deviations greater than 5 mm, the affected bosses and surfaces were ground and refinished to the required dimensions. In the later stages of development, I focused on controlling the vacuum more tightly, increasing the sand compaction by using a vibrating table with adjustable frequency and amplitude, and slightly lowering the pouring temperature for ductile iron. These changes reduced the frequency of width oversize. The relationship between the vacuum level and the allowable mold wall displacement can be approximated by a simple equation:
$$ \Delta w = \frac{F_{expand} – F_{vacuum}}{K_{sand}} $$
where $\Delta w$ is the dimensional deviation, $F_{expand}$ is the expansion force of the solidifying metal, $F_{vacuum}$ is the restraining force provided by the vacuum-induced pressure difference, and $K_{sand}$ is the effective stiffness of the sand mass. By increasing the vacuum from 0.05 to 0.06 MPa, $F_{vacuum}$ increases proportionally, thereby reducing $\Delta w$. Similarly, a lower pouring temperature reduces $F_{expand}$ because the solidification time shortens and the ferrostatic pressure is lower. The sand stiffness can be improved by using finer sand or increasing vibration time, but this must be balanced against sand permeability.
One of the most valuable observations I made was that the bosses around the cooling water inlet and outlet pipes, as well as the bolt hole bosses, exhibited far better dimensional accuracy in lost foam castings than in traditional green sand or resin sand molds. This is because the foam pattern itself defines the exact contour of these features, and the refractory coating transfers the pattern dimension with high fidelity. In contrast, sand molds made from patterns depend on draft angles and core shifts. The surface finish of the lost foam castings cooling walls was also clearly superior, with no visible parting lines, sand inclusions, or core finning. This reduced the finishing time significantly.
Economic Comparison of Casting Processes
The economic advantage of lost foam castings for cooling walls was one of my primary motivations. I performed a detailed cost analysis based on one row of cooling walls, which consisted of 20 castings each weighing 1.7 tons. I compared the molding-related costs for three processes: lost foam, resin sand, and sodium silicate (water glass) sand. I excluded the cost of the embedded steel tubes and melting because those are identical regardless of mold process. The sand cost was also nearly the same because both resin sand and lost foam have high reclamation rates of up to 95%. However, the binder costs, labor costs, and pattern costs differed greatly.
For the lost foam castings process, the cost components were the foam pattern material and the labor for foam pattern assembly. The foam pattern cost was 60 CNY per ton of castings, and the labor was 390 CNY per ton. Thus, the total molding cost for the row was:
$$C_{LFC} = 1.7 \times 20 \times (60 + 390) = 15{,}300 \text{ CNY} $$
For the resin sand process, the resin price was 13,500 CNY per ton, and the hardener was 4,000 CNY per ton. Each ton of castings consumed 48 kg of resin and 25 kg of hardener. Additionally, coating, nails, and parting compound cost 45 CNY per ton, and the wooden pattern cost 4,100 CNY with an expected life of 60 castings (i.e., three batches of 20). Thus, the cost was:
$$C_{resin} = 1.7 \times 20 \times (48 \times 13.5 + 25 \times 4) + 1.7 \times 20 \times 45 + \frac{4{,}100}{3} = 28{,}328.67 \text{ CNY} $$
For the sodium silicate (water glass) sand process, the new sand consumption was 1,000 kg per ton of castings at 200 CNY per ton, water glass consumption was 90 kg per ton at 3,800 CNY per ton, auxiliary consumables were 19 CNY per ton, and CO₂ gas for hardening consumed 3 bottles per ton at 23 CNY per bottle. The wooden pattern cost was again 4,100 CNY over three batches. The total cost was:
$$C_{silicate} = 1.7 \times 20 \times (200 + 90 \times 3.8) + 1.7 \times 20 \times 19 + \frac{4{,}100}{3} + 1.7 \times 20 \times 3 \times 23 = 22{,}786.67 \text{ CNY} $$
Table 3 summarizes the comparison. The lost foam castings process provided a savings of more than 13,000 CNY compared with resin sand, and more than 7,400 CNY compared with sodium silicate sand, for just a single row of cooling walls. Given that our foundry often produced multiple rows per month, the annual savings were substantial.
| Cost item | Lost foam castings | Resin sand | Sodium silicate sand |
|---|---|---|---|
| Foam pattern material | 60 CNY/t | – | – |
| Foam assembly labor | 390 CNY/t | – | – |
| Resin + hardener cost | – | 48×13.5 + 25×4 CNY/t | – |
| Water glass cost | – | – | 90×3.8 CNY/t |
| New sand cost | ~0 (95% reclamation) | ~0 | 200 CNY/t |
| Auxiliary consumables | included | 45 CNY/t | 19 CNY/t |
| CO₂ gas | – | – | 3×23 CNY/t |
| Pattern amortization | not needed (foam pattern) | 4,100/3 CNY | 4,100/3 CNY |
| Total cost | 15,300 CNY | 28,328.67 CNY | 22,786.67 CNY |
In addition to the direct cost savings, lost foam castings significantly reduced the time and labor for cleaning and finishing. Because the foam pattern leaves no parting lines and the sand is unbonded, the castings came out with a clean surface requiring only shot blasting and minor grinding at the gating areas. For resin sand or sodium silicate sand castings, we had to perform extensive knockout, core removal (for the embedded tube portions), and grinding of fins. The lost foam castings cooling walls required only a minor cleaning of the bolt holes and the pipe openings. This accelerated our production lead time and improved the overall throughput of the foundry.
Metallurgical Quality of Lost Foam Castings
Many foundry engineers worry that the pyrolysis of the foam pattern might introduce carbon or gas-related defects into the casting. In our production of cooling walls, I carefully monitored the carbon and gas content of the iron. For gray iron HT200, the carbon content in the final product was 3.44–3.48%, which is within the specified range. For ductile iron, the carbon was 3.59–3.81%. The sulfur and phosphorus levels were also as expected. No consistent increase in carbon due to foam decomposition was observed because the refractory coating effectively prevented direct contact between the metal and the foam residue. The coating acts as a barrier that allows gas to escape while keeping solid carbon particles from entering the metal.
The mechanical properties listed in Table 2 show that the tensile strength of HT200 exceeded 200 MPa in every heat, with values of 206 and 211 MPa. For QT500-7, the tensile strengths were 481 and 474 MPa, and the elongations were 5.5% and 5.9%, comfortably above the minimum of 4%. QT400-18 achieved tensile strengths of 440 and 456 MPa with elongations of 13.9% and 14.6%, far exceeding the specified 12%. I attribute these high elongations to the clean microstructure obtained through proper inoculation and the relatively slow cooling rate in the thick cooling wall sections, which promoted fully ferritic matrices. The vacuum conditions in lost foam castings also help in producing a sound casting without internal shrinkage porosities, because the negative pressure assists in feeding the liquid metal through the mold cavity.
For a quantitative relationship between the cooling rate and the nodule count, I used the following empirical formula:
$$N_v = K \cdot \frac{1}{\sqrt[3]{t_s}} \cdot \exp\left(-\frac{T_{pour} – T_{liquidus}}{T_0}\right)$$
where $N_v$ is the nodule count per square millimeter, $t_s$ is the local solidification time in seconds, $T_{pour}$ is the pouring temperature, $T_{liquidus}$ is the liquidus temperature of the alloy, and $K$ and $T_0$ are constants related to the inoculant efficiency. In our lost foam castings, the solidification time was slightly longer than in sand casting because of the insulating effect of the unbonded sand and the vacuum, but the nodule count remained in the range of 100–150 per mm², which is considered excellent for these grades.
Process Control and Optimization
To ensure consistent quality of the lost foam castings, I developed a process control checklist that covered every critical parameter. The foam pattern density of 20–22 g/L was selected because lower density (less than 18 g/L) caused pattern collapse during coating application, while higher density (above 25 g/L) produced more gas and increased the risk of carbon defects. The coating viscosity was measured every batch using a Ford cup, and the coating thickness was verified with a wet film gauge. The drying conditions were maintained with an air-circulated oven at 50 °C for 8 hours. The sand used was silica sand with an AFS grain fineness of 40–45. The sand was dried to a moisture content below 0.2% before the flask filling. During flask filling, the sand was vibrated at a frequency of 50 Hz and an amplitude of 0.5 mm for 90 seconds to achieve a bulk density of 1.55–1.60 g/cm³. The vacuum was calibrated using a differential pressure gauge connected to the flask outlet. The pouring rate was controlled by a stopper rod system with a tundish to ensure a steady stream, which minimizes turbulence and prevents the metal from prematurely bridging the foam pattern.
One of the unique challenges of large cooling walls in lost foam castings is the possibility of pattern deformation during sand filling. The pattern, with its large planar dimensions and multiple protruding bosses, must resist the lateral forces from the flowing sand. I addressed this by adding internal foam ribs that act as spacers, and by placing the pattern on a steel support grid inside the flask. The ribs were removed during the finishing stage. These measures helped to keep the width dimension within the acceptable tolerance in most castings.
The gating system for lost foam castings also required careful design. I used a bottom-gating system with a ceramic foam filter to trap inclusion particles and promote laminar flow. The gating ratio (sprout:runner:ingate) was set as 1:1.5:1.2 for gray iron and 1:2:1.2 for ductile iron. Because the foam pattern degrades to gas, the ingates were placed at locations where the gas could escape through the coating quickly. The pouring basin was designed with a stopper to control the initial flow rate. Using the Bernoulli equation as a guide, I calculated the ideal pouring time as:
$$t_{pour} = \frac{2 \cdot W}{\rho \cdot A_g \cdot \sqrt{2 g h}}$$
where $W$ is the total weight of liquid metal poured, $\rho$ is the density of the molten iron, $A_g$ is the total cross-sectional area of the ingates, $g$ is the gravitational acceleration, and $h$ is the effective pouring head height. For our cooling walls, the computed pouring time was 120–150 seconds, which I maintained closely by adjusting the stopper rod opening.
Defects and Remedies in Lost Foam Castings
No manufacturing process is without defects, and lost foam castings are no exception. During the initial trials, I encountered two main defects: surface folds and gas porosity. Surface folds appeared as thin, irregular wrinkles on the upper surfaces of the cooling wall. This was caused by the carbon residue from the foam pattern being insufficiently flushed away. I remedied this by increasing the vacuum from 0.04 to 0.05 MPa for gray iron and by raising the pouring temperature by 10–20 °C to allow more complete decomposition and evacuation of the foam gas. Additionally, I switched to a coating with higher permeability, which allowed the gas to escape more rapidly.
Gas porosity was detected in the first ductile iron cooling wall cast as a lost foam casting. The porosity was concentrated in the thick sections near the bolt holes. I analyzed the gas pressure evolution and determined that the vacuum level was too low for the volume of gas generated. After raising the vacuum to 0.06 MPa and adding vent holes through the foam pattern (small cylindrical foam pieces that connect the pattern interior to the sand), the porosity disappeared. The vent holes acted as preferential gas escape channels, reducing the gas pressure in the liquid metal during solidification.
Another potential defect in lost foam castings is sand penetration, where the molten metal seeps between the coating and the sand. This was prevented by ensuring the coating thickness was at least 1.0 mm and by using a finer sand. I also used a more refractory coating for the bottom faces where the metal pressure is highest. After these changes, the surface quality of all subsequent cooling walls was excellent, with no sand adhesion.
Environmental and Sustainability Aspects
The lost foam castings process offers significant environmental benefits compared with conventional sand molding. In my production department, the sand reclamation rate reached 95%. Only a small fraction of the sand, typically the fines and the sand that adheres to the coating, was discarded. In contrast, sodium silicate sand had a reclamation rate below 60% because the sodium silicate binder forms a hard film on the sand grains that cannot be easily removed. The lower binder consumption and the absence of resin also meant that no toxic fumes were released during pouring, except for the harmless pyrolysis products of EPS, which were collected by the vacuum system and filtered. This reduced the foundry’s solid waste and air emissions. The thermal decomposition of the foam pattern produces only carbon dioxide and a small amount of carbon black, which is much more benign than the sulfur and phenolic compounds released from some resin binders.
In terms of energy consumption, lost foam castings require less energy for shakeout and cleaning because the unbonded sand falls away from the casting. We also eliminated the need for core making because the foam pattern already includes the internal geometries. This shortened the overall production cycle from nearly 10 days to 4 days for a row of cooling walls. The productivity gain was especially important during peak production periods.
Conclusion and Future Perspectives
Through my experience, I have demonstrated that the lost foam castings process is not only technically feasible for manufacturing blast furnace cooling walls but also economically advantageous. The castings produced by lost foam castings consistently met the chemical composition, metallurgical structure, and mechanical property requirements of both gray iron and ductile iron specifications. The dimensional accuracy of the critical boss features was superior to that achieved by conventional sand casting, and the surface quality was visibly better. The overall production cost, when compared with resin sand and sodium silicate sand processes, was significantly lower—13,000 CNY and 7,400 CNY savings per row, respectively. Moreover, the lost foam castings process reduced solid waste, lowered energy consumption, and improved the productivity of the foundry.
However, there remained a minor issue: an occasional width oversize of 6–10 mm in some ductile iron castings. By refining the vacuum control, optimizing the sand compaction, and lowering the pouring temperature, I was able to minimize this deviation but not entirely eliminate it. Further improvements could be made by using a more rigid sand container or by applying a higher compaction force. In the future, I believe that lost foam castings will become the preferred method for producing complex cored castings in heavy iron foundries, not only for cooling walls but also for engine blocks, cylinder heads, and pipe fittings. The ability to combine multiple features into a single foam pattern, with no draft angles, no cores, and no parting lines, fundamentally changes the economics of casting design. As we continue to optimize the process through computational simulation and sensor-controlled vacuum systems, the remaining dimensional challenges will be resolved, and lost foam castings will achieve even wider adoption.
The decisive factors for successful lost foam castings in my experience are: the use of high-quality foam patterns with low density and high structural integrity, a robust refractory coating with controlled permeability, a tightly controlled vacuum level, and a well-designed gating system that ensures smooth filling and adequate gas evacuation. When these elements are properly managed, lost foam castings deliver excellent performance and reliability. I strongly recommend that foundries considering this process for new products first run a detailed feasibility study and trial castings, as the process does have a learning curve. But once mastered, the rewards in cost, quality, and environmental stewardship are immense.
In summary, the journey of adopting lost foam castings for blast furnace cooling walls was challenging but extremely rewarding. Every cooling wall we shipped validated our confidence in the process. The advantages of lost foam castings in terms of sand reclamation, environmental friendliness, dimensional accuracy, and cost efficiency make it one of the most promising casting methods for the future. I believe that by sharing my detailed insights, other foundry engineers can accelerate their own process development and avoid the pitfalls I encountered. The key is to embrace the unique behavior of the foam pattern in the mold and to diligently control every step from pattern making to final inspection. With these practices, lost foam castings will continue to lead the way toward cleaner and more efficient foundry production.